One year of basic income in Minneapolis

One year of basic income in Minneapolis
One year of basic income in Minneapolis

By Andrew Goodman-Bacon, Ryan Nunn, and Vanessa Palmer

See original post here.

A wave of experimentation in the way that U.S. cities, counties, and states support households with low incomes is sweeping the country, and Minneapolis is at the leading edge. While the backbone of America’s safety net—public health insurance, food benefits, and cash or social insurance for specific groups—remains in place, over 100 localities are testing what happens when low-income people are simply provided with income, no strings attached. Minneapolis’ guaranteed basic income (GBI) pilot program, initiated by city officials in 2021, uses federal pandemic support funds to give $500 a month to 200 low-income households for two years. One year in, GBI is creating measurable improvements in people’s lives.

One year in, GBI is creating measurable improvements in people’s lives.

“People know where they need to spend,” Minneapolis Mayor Jacob Frey told MPR News in 2022, arguing that GBI would let low-income people use the resources as they see fit, which has the potential to help them more than existing programs that require benefits to be spent on specific goods like food, rent, or utilities.

One year after GBI payments began, an evaluation conducted by our team of researchers at the Federal Reserve Bank of Minneapolis shows that the randomly selected recipients have better mental health, more stable finances, and higher food security than households with the same poverty status and from the same areas of the city who did not receive payments. Recipients spend their GBI money—which represents about a one-third boost to the total annual income of a typical participant—on major monthly expenses like rent and food. We do not find evidence that payments cause recipients to work less, a common concern about GBI programs.

GBI is a new type of program

There are important differences between GBI and existing U.S. income support programs. For example, one of the largest cash transfers in the U.S. is the earned income tax credit (EITC), which delivers a one-time annual payment of up to $7,430 to low-income families. This amount is comparable to the $6,000 a year that Minneapolis GBI recipients receive. Unlike GBI, however, the EITC goes only to those with earned income, which means that it does little for families with the lowest incomes, including long-term unemployed workers. GBI, on the other hand, has no work requirement. In that way, it provides more financial support for the poorest households but raises questions about whether it leads recipients to work less.

Basic income also differs from traditional safety net programs in its simplicity. The Supplemental Nutrition Assistance Program, for example, requires recipients to spend their benefits on certain goods at certain stores to encourage them to eat specific types of food. GBI does neither of these things, giving people more flexibility and making the program easier to administer while making no attempt to channel or restrict their purchases.

It is also important to differentiate targeted basic income programs, like Minneapolis’ pilot, from universal programs. The latter would go to everyone, regardless of their income or other characteristics, and is usually thought of as a response to a potentially broader problem than existing poverty. For example, former head of the Service Employees International Union Andy Stern has argued for a universal basic income on the grounds that automation may someday eliminate many more jobs than it creates. In Minneapolis, however, the basic income pilot program was available only to low-income households in low-income ZIP codes in the central and northern parts of the city.

So how might unrestricted income support for a target group of Minneapolis households work? Would it address short-term economic instability or shift households’ longer-term trajectories? Perhaps just as important, would any estimated effects of the GBI program warrant its further consideration as a permanent policy? These are just the kinds of questions the City of Minneapolis set out to answer when it selected the Fed to evaluate its GBI pilot.

The Minneapolis GBI pilot

With its goals of understanding the economy and supporting the prosperity of low- and moderate-income communities, the Fed has a clear interest in the neutral evaluation of a policy like GBI. The City of Minneapolis determined its policy objectives, made program design decisions, and administered the pilot. Over the last two years, our Fed team has worked closely with the city to design a rigorous pilot study and analyze the resulting confidential data.

We first developed an extensive survey to gather information on the kinds of outcomes GBI might change. These include labor market behaviors, self-assessed economic precarity, food and housing security, and psychological health. Most of the survey comprises questions from preexisting surveys so that we can compare our results with those of other studies.

The biggest challenge in evaluating whether GBI payments cause important changes in recipients’ lives is creating groups that are similar in all respects except that one gets GBI and the other does not. This is where the “randomized” part of our trial helps. From an initial pool of 530 applicants who filled out baseline survey information on their economic activities and life circumstances, the city randomly chose 200 to receive $500 monthly GBI payments. The other 230 applicants serve as the comparison group and receive small payments for answering surveys but no other financial assistance. Because the only thing that determined whether an individual would or would not get GBI payments was a random number, recipients’ baseline survey answers about household income, family size, employment, food security, financial stability, or psychological distress were no different between either group.

As city officials worked to enroll study participants, however, it became clear that only about half of the initial comparison group would continue to participate. This created a challenge for our evaluation. Which participants were opting out? If they differed from those who opted in—if they were more likely to lose jobs, to move, or to have mental health crises, for example—then comparing them with the payment participants would be misleading.

Fortunately, our collaboration with the city uncovered this issue early and we settled on a set of statistical methods to try to address it. We posted this “pre-analysis plan” to a clinical trial registry before we analyzed the six-month data, which ensured that none of our analytical choices were driven by after-the-fact data anomalies. In keeping with the Fed’s mission to serve the public interest, our research plan is part of a suite of high-quality openly shared resources meant to help guide GBI studies still under development and to facilitate comparisons with GBI studies focusing on different groups, cities, or policy details.

What have we learned?

The City of Minneapolis pilot is still underway, but we now have results from six- and 12-month follow-up surveys. We focused our analysis on six domains, each of which bundles several related questions into one measure. For example, our “labor supply index” combines responses about employment, job searching, full-time or part-time status, multiple jobs, and usual weekly hours. Combining these questions helps us detect smaller GBI effects than would a single question alone, and speaks to the broad concept of how GBI recipients navigate the labor market.

We take a similar approach with five other core indexes. The housing stability index includes questions about things like overcrowding and forced moves. The financial security index includes subjective questions about people’s general financial status and their ability to afford expenses, as well as objective questions about borrowing. The well-being index includes questions about life satisfaction. The food security index includes questions about food adequacy and frequency of going without meals. Finally, the psychological wellness index comes from a clinically validated depression scale.

For some but not all of these indexes, we find clear evidence of a GBI effect. The figure shows average outcomes at baseline, six months after payments began, and one year after payments began for participants who responded to our 12-month survey. The charts show outcome changes for payment group members, who received $6,000 as of the 12-month survey, and outcome changes for survey group members, who did not receive GBI payments.

Food security, financial security, well-being, and psychological wellness all improved for GBI recipients relative to the comparison group of nonrecipients as of the 12-month mark. Additionally, we find some evidence of a positive impact on housing stability. For the remaining index—labor supply—we cannot detect a significant impact of GBI. Note that these changes in average outcomes do not necessarily correspond to the formal effects that we estimate using methods intended to adjust for survey nonresponse. More details can be found in the cumulative results of our evaluation of the pilot program as well as in our pre-analysis plan.

After one year of payments, GBI program shows positive impacts for recipients on several indicators.

Source note: The “payment group” received Minneapolis guaranteed basic income pilot program payments; the “survey group” did not receive basic income payments. For visual comparability, the results shown here reflect the responses over time from the participants who responded to the 12-month survey and who had been verified to be study-eligible as of that time. The analytical results discussed in the text include all eligible survey responses.
Source: Authors’ calculations using original survey data.

The story emerging from our Minneapolis study matches the few available findings from other GBI pilots quite closely. The most well-known basic income pilot—and one of the few others with public results—is from Stockton, California. The analysis of that program found improvements in psychological well-being nearly identical to our findings, and no reductions in employment. A GBI pilot that operated from November 2020 through August 2021 serving a low-income immigrant community in Chelsea, Massachusetts, also showed that GBI lowered financial distress but not labor supply. The immediate improvements in food security in that study, however, did not last like ours did.

All of these pilots have unique features: They reach different target populations, have differing levels of support and duration, and their outcomes are measured in different ways. Collectively, they will build a much richer understanding of how basic income matters to low-income Americans.

What do we not know yet?

We are still in the early days of basic income programs, both nationally and in Minneapolis. There are at least four areas to consider as the conversation moves forward.

First, there are many lessons yet to be learned from our Minneapolis pilot. A complete picture of how the Minneapolis GBI pilot affects short-run outcomes for participants will still require another two years to become clear. We will continue to collect and analyze data through the end of GBI payments in 2024, during which time our results could evolve as people’s choices and outcomes do.

Our results could evolve as people’s choices and outcomes do.

Second, there are many lessons yet to be learned from other GBI studies. With much experimentation happening in such different ways, it will be crucial to synthesize what we learn from all available studies. One reason for the transparency of our research methods is to make these comparisons as accurate as possible.

Third, there are important lessons that cannot be learned from small pilots—even high-quality randomized ones. Were the City of Minneapolis, the State of Minnesota, or the federal government to launch a basic income program, vastly more participants would be enrolled and effects on local, state, or national communities could be quite different. People could respond differently to a permanent program than they do to a time-limited pilot. Research on large, permanent programs that resemble GBI, such as cash transfers from the Alaska Permanent Fund or from tribal gaming revenues, generally agree with our findings that GBI improves self-assessed well-being and stability without reducing labor supply. Nevertheless, policy experimentation on a larger scale would help us better understand if and how actual programs would function differently than pilots.

And fourth, we don’t yet know the effects of financing a permanent GBI program. Simple calculations suggest that a national program would be very expensive. Therefore, while local pilots can tell us something about how GBI recipients respond to payments, they cannot tell us how people who fund a GBI program—either through taxes or through offsetting reductions in existing transfer programs—would respond.

Why does any of this matter?

The Federal Reserve Bank of Minneapolis is using rigorous methods to evaluate the Minneapolis pilot because it sees value in helping policymakers better understand an emerging program that (1) matters for low- and moderate-income communities and (2) could contribute in important ways to how labor markets function.

Evidence derived from pilots will be critical for policymakers to make informed decisions about any permanent implementation of a basic income.

Basic income programs are currently the subject of considerable discussion, both in our local communities and throughout the country. While at the moment such programs exist only on a temporary basis, advocates aim to make them a permanent feature of the social safety net. Evidence derived from pilots will be critical for policymakers to make informed decisions about any permanent implementation of a basic income.

You may also be interested in...


  • This field is for validation purposes and should be left unchanged.

Join our community and sign up for the Basic Income Today newsletter.

  • This field is for validation purposes and should be left unchanged.